Modeling and Simulation of Vehicular Networks: Towards ...cdn.intechweb.org/pdfs/12878.pdf · VANET...

1. Introduction

Vehicular Ad Hoc Networks (VANETs) have been envisioned with three types of applicationsin mind: safety, traffic management, and commercial applications. By using wirelessinterfaces to form an ad hoc network, vehicles will be able to inform other vehicles abouttraffic accidents, hazardous road conditions and traffic congestion. Commercial applications(e.g., data exchange, audio/video communication) are envisioned to provide incentive forfaster adoption of the technology.To date, the majority of VANET research efforts have relied heavily on simulations, due toprohibitive costs of employing real world testbeds. Current VANET simulators have gone along way from the early VANET simulation environments, which often assumed unrealisticmodels such as random waypoint mobility, circular transmission range, or interference-freeenvironment (Kotz et al. (2004)). However, significant efforts still remain in order to enhancethe realism of VANET simulators, at the same time providing a computationally inexpensiveand efficient platform for performance evaluation of VANETs. In this work, we distinguishthree key building blocks of VANET simulators:

– Mobility models,

– Networking (data exchange) models,

– Signal propagation (radio) models.

Mobility models deal with realistic representation of vehicular movement, including mobilitypatterns (i.e., constraining vehicular mobility to the actual roadway), interactions between thevehicles (e.g., speed adjustment based on the traffic conditions) and traffic rule enforcement(e.g., intersection control through traffic lights and/or road signs). Networking models aredesigned to provide realistic data exchange, including simulating the medium access control(MAC) mechanisms, routing, and upper layer protocols. Signal propagation models aim atrealistically modeling the complex environment surrounding the communicating vehicles,including both static objects (e.g., buildings, overpasses, hills), as well as mobile objects (othervehicles on the road).We first present the state-of-the art in vehicular mobility models and networking modelsand describe the most important proponents for these two aspects of VANET simulators.Then, we describe the existing signal propagation models and motivate the need for more

Mate Boban1,2, Tiago T. V. Vinhoza2

1Department of Electrical and Computer Engineering, Carnegie Mellon University5000 Forbes Avenue, Pittsburgh, PA, 15213, USA

2Instituto de Telecomunicacoes, Departamento de Engenharia Electrotecnica e de ComputadoresFaculdade de Engenharia da Universidade do Porto, 4200-465, Porto, Portugal

USA and Portugal

Modeling and Simulation of Vehicular Networks: Towards Realistic and Efficient Models

8

www.intechopen.com

2 Theory and Applications of Ad Hoc Networks

accurate models that are able to capture the behavior of the signal on a per-link basis,rather than relying solely on the overall statistical properties of the environment. Morespecifically, as shown in (Koberstein et al. (2009)), simplified stochastic radio models (e.g.,free space (Goldsmith (2006)), log-distance path loss (Rappaport (1996)), two-ray groundreflection (Goldsmith (2006)), etc.), which are based on the statistical properties of the chosenenvironment and do not account for the specific obstacles in the region of interest, areunable to provide satisfactory accuracy for typical VANET scenarios. Contrary to this,topography-specific, highly realistic channel models (e.g., based on ray tracing (Maurer et al.(2004))) yield results that are in very good agreement with the real world. However, thesemodels are computationally too expensive and usually bound to a specific location (e.g., aparticular neighborhood in a city), thus making them impractical for extensive simulationstudies. For these reasons, such models are not implemented in VANET simulators. Basedon the experimental assessment of the impact of mobile obstacles on vehicle-to-vehiclecommunication, we point out the importance of the realistic modeling of mobile obstacles andthe inconsistencies that arise in VANET simulation results in case these obstacles are omittedfrom the model. Motivated by this finding, we developed a novel model for incorporating themobile obstacles (i.e., vehicles) in VANET channel modeling. A useful model that accountsfor mobile obstacles must satisfy a number of requirements: accurate vehicle positioning,realistic underlying mobility model, realistic propagation characterization, and manageablecomplexity. The model we developed satisfies all of these requirements (Boban et al. (2011)).The proposed model accounts for vehicles as three-dimensional obstacles and takes intoaccount their impact on the LOS obstruction, received signal power, and the packet receptionrate. The algorithm behind the model allows for computationally efficient implementation inVANET simulators. Furthermore, the proposed model can easily be used in conjunction withthe existing models for static obstacles to accurately simulate the entire spectrum of VANETenvironments with regards to both road conditions (e.g., sparse or dense vehicular networks),as well as various surroundings (including highway, suburban, and urban environments).

VANET Simulation

Environment

Data Exchange

Modeling

Signal Propagation

ModelingMobility Modeling

Traffic Rule Enforcement

(intersection

management, speed

modeling, etc.)

Vehicle Interaction

Models (lane changing,

car following, accident

simulation, etc.)

Stochastic ModelsDeterministic Models

Static Obstructions

Modeling (e.g., road

surface, buildings, hills,

foliage, etc)

Mobile Obstructions

Modeling (moving

vehicles)

Trace-based Mobility

Models

Real-world Traces Artificial Traces

Dedicated Traffic

Models

Mobility Patterns

(random waypoint,

Manhattan grid, road-

constrained, etc.)

Fig. 1. Structure of VANET simulation environment

2. Mobility models

Mobility models can be roughly divided in trace-based models and dedicated traffic models(Fig. 1). Trace-based models are based on a set of generated vehicular traces which are then

42 Mobile Ad-Hoc Networks: Applications

www.intechopen.com

Modeling and Simulation of Vehicular Networks: Towards Realistic and Efficient Models 3

used as an underlying mobility pattern over which the data communication is carried over.The traces can be either real world (i.e., based on mapping of the positions of vehicles)(Ferreira et al. (2009) and Ho et al. (2007)), or artificially generated using the dedicatedtraffic engineering tools (Naumov et al. (2006)). The advantage of trace-based models is theyprovide the highest level of realism achievable in VANET simulations. However, there arealso several important shortcomings. Firstly, in order to collect the real world mobility traces,significant time and cost are involved. This often makes the traces collected limited withrespect to both the number of the vehicles that are recorded and the region over which therecording has been made. Further this implies that there is rarely a chance to record themobility of all the vehicles in a certain region (as it would often involve equipping eachvehicle with the location devices), thus leading to a need for compensating algorithms forthe non-recorded vehicles. Finally, since the traces are collected/recorded beforehand, thefeedback connection from the networking model to the mobility model is not available.This is a very important shortcoming, given that a large number of proposed IntelligentTransportation System (ITS) applications carried over VANETs can affect the movement of thevehicles (this is especially the case with traffic management applications), thus rendering thetrace-based models inadequate for any application with the feedback loop between the trafficand networking models. A vivid example of such application is Congested Road Notification(Bai et al. (2006)), which aids the vehicles in circumventing congested roads, thus directlyaffecting the mobility of the vehicles through the network communication.A characteristic that distinguishes the dedicated traffic models from the trace-based ones,capability to support the feedback loop between the mobility model and the networkingmodel, is an important reason for adopting the more flexible dedicated traffic models. Thisway, the information from the networking model (e.g., a vehicle receives a traffic updateadvising the circumvention of a certain road) can affect the behavior of the mobility model(e.g., the vehicle takes a different route than the one initially planned). Early VANET mobilitymodels were characterized by their simplicity and ease of implementation. For quite sometime, the random waypoint mobility model (Saha & Johnson (2004)), where the vehicles moveover a plane from one randomly chosen location to another, was the de facto standard forVANET simulations. However, it was shown that the overly simplified mobility models suchas random waypoint are not able to model the vehicular mobility adequately (Choffnes &Bustamante (2005)). A significant step towards the realism were the simple one-dimensionalfreeway model and the so-called Manhattan grid model (Bai et al. (2003)), where the mobilityis constrained to a set of grid-like streets which represent an urban area. Further elaborationof the mobility models was achieved by using map generation techniques, such as Voronoigraphs (Davies et al. (2006)), which constrain the movement of the vehicles to a network ofartificially generated irregular streets. Recently, the most prominent mobility models (e.g.,Choffnes & Bustamante (2005) and Mangharam et al. (2005)) started utilizing real world mapsin order to constrain the vehicle movement to real streets based on some of the availablegeospatial databases (e.g., the U.S. Census Bureau’s TIGER data (U.S. Census Bureau TIGERsystem database (n.d.)) or the data collected in the OpenStreetMap project (Open Street MapProject (n.d.))). Furthermore, the distinction can be made with regards to the coupling betweenthe mobility and networking and signal propagation models. On one side of the spectrumare the mobility models embedded with the networking model (Choffnes & Bustamante(2005)), which allow for a more efficient execution of the simulation. On the other side,there are mobility models which are based on the dedicated traffic simulators stemming fromthe traffic engineering community (e.g., SUMO - Simulation of Urban MObility (n.d.) and

43Modeling and Simulation of Vehicular Networks: Towards Realistic and Efficient Models

www.intechopen.com


CORSIM: Microscopic Traffic Simulation Model (n.d.)), which are then bidirectionally coupledwith the networking model (e.g., Piorkowski et al. (2008)). These types of environmentsare characterized by a high level of traffic simulation credibility, but often suffer frominefficiencies caused by the integration of two separate systems (Harri (2010)).Vehicle interaction (Fig. 1) includes modeling the behavior of a vehicle that is a directconsequence of the interaction with the other vehicles on the road. This includes themicroscopic aspects of the impact of other vehicles, such as lane changing (Gipps (1986)) anddecreasing/increasing the speed due to the surrounding traffic, as well as the macroscopicaspects, such as taking a different route due to the traffic conditions (e.g., congestion). Anotherimportant aspect of mobility modeling is traffic rule enforcement, which includes intersectionmanagement, changing the vehicle speed based on the speed limits of the roads, and generallymaking the vehicle obey any other traffic rules set forth on a certain highway. Even thoughthe vehicle interaction and the enforcement of traffic rules were shown to be essential foraccurate modeling of vehicular traffic (Helbing (2001)), as noted in (Harri (2010)), manyof VANET mobility models have scarce support for these microscopic aspects of vehicularmobility. For this reason, significant research efforts remain in order to make these aspects ofmobility models more credible, and for the research community to strive for the simulationenvironments that realistically model these components.With regards to the implementation approaches for the mobility models, the most prolificproponents are (Helbing (2001)): the cellular automata models (Nagel & Schreckenberg(1992)), the follow-the-leader models (e.g., car-following (Rothery (1992)) and intelligentdriver model (Treiber et al. (2000))), the gas-kinetic models (Hoogendoorn & Bovy (2001)),and the macroscopic models (Lighthill & Whitham (1955)). Further classification of mobilitymodels can be made with respect to the granularity at which the mobility is simulated,categorizing the mobility models as microscopic, mesoscopic, and macroscopic. Microscopicmodels are simulating the mobility at the per-vehicle level (i.e., each vehicle’s motion issimulated separately). Prominent examples of such models are the car following model(Rothery (1992)) and cellular automata models (Nagel & Schreckenberg (1992) and Tonguzet al. (2009)). Macroscopic models simulate the entire vehicular network as an entity thatpossesses certain physical properties. Such models can give insights into the overall statisticalproperties of vehicular networks (e.g., the average vehicular density, average speer, or theflow/density relationship of a given vehicular network). Examples of such models arekinematic wave models (Jin (2003)) and fluid percolation (Cheng & Robertazzi (Jul. 1989)).Mesoscopic models are simulating the mobility at the flow level, where a number of vehiclesis characterized by certain averaging properties (e.g., arrival time, average speed, etc.), but theflows are distinguishable. Gas-kinetic model (Hoogendoorn & Bovy (2001)) is an example ofmesoscopic models. For an extensive treatment focusing on modeling the vehicular traffic ingeneral, we refer the reader to (Helbing (2001)), and for the overview of the mobility modelsused in VANET research, we refer the reader to (Harri (2010)).

3. Networking models

Unlike the mobility models or signal propagation models for VANETs, which have significantdifferences when compared to models used in other types of mobile ad hoc networks(MANETs) (Murthy & Manoj (2004)), the networking models for VANETs are quite similar tothose used in other fields of MANET research. The data models used in the current simulators,such as NS-2 (Network Simulator 2 (n.d.)), JiST/SWANS/STRAW (Choffnes & Bustamante(2005)), and NCTU-NS (Wang et al. (2003)), rely on discrete event simulation, where different


www.intechopen.com


protocols of the network stack are executed based on the events triggered either by upperlayer (e.g., an application sends a message to the networking protocol) or by lower layer(e.g., the link layer protocol notifies the network layer protocol about the correct receptionof the message). The main difference arises in the use of a dedicated VANET protocol stackcalled Wireless Access in Vehicular Environments (WAVE), standardized under the IEEE 1609working group (IEEE Trial-Use Standard for Wireless Access in Vehicular Environments (WAVE) -Networking Services (Apr. 2007)).In 1999, the U.S. Federal Communications Commission (FCC) allocated 75 MHz of spectrumbetween 5850 - 5925 MHz for WAVE systems operating in the Intelligent TransportationSystem (ITS) radio service for vehicle-to-vehicle (V2V) and infrastructure-to-vehicle (V2I)communications. Similarly, the European Telecommunications Standards Institute (ETSI) hasallocated 30 MHz of spectrum in the 5.9 GHz band for ITS services in August 2008, andmany other countries are actively working towards standardizing the 5.9 GHz spectrum,thus allowing worldwide compatibility of WAVE devices in the future. WAVE provisionsfor public safety and traffic management applications. Commercial (tolling, comfort (Bai et al.(2006)), entertainment (Tonguz & Boban (2010)), etc.) services are also envisioned, creatingincentive for faster adoption of the technology. The lower layers of the WAVE protocolstack are being standardized under the Dedicated short-range communications (DSRC) setof protocols (IEEE Draft Standard IEEE P802.11p/D9.0 (July 2009)). DSRC is based on IEEE802.11 technology and is proceeding towards standardization as IEEE 802.11p. Fig. 2 showsthe WAVE protocol stack. On the network layer, WAVE Short Message Protocol (WSMP)is being developed for fast and efficient message exchange in VANETs. It is planned tosupport both safety as well as for non-safety applications. Applications running over WSMPwill directly control the physical layer characteristics (e.g., channel number and transmitterpower) on a per message basis. As seen in Fig. 2, applications running over the standardTCP/IP protocol stack are also supported. Their operation is restricted to the predefinedunderlying physical layer characteristics, based on the application type. The applicationswill be divided in up to eight levels of priority, with the safety applications having the highestlevel of priority. The multi-channel operation (IEEE Trial-Use Standard for Wireless Access inVehicular Environments (WAVE) - Multi-channel Operation (2006)) is aimed at providing higheravailability and managing contention. Channels are divided into Control Channel (CCH)and Service Channels (SCH). WAVE devices must monitor the Control Channel (CCH) forsafety application advertisements during specific intervals known as CCH intervals. CCHintervals and are specified to provide a mechanism that allows WAVE devices to operate onmultiple channels while ensuring all WAVE devices are capable of receiving high-prioritysafety messages with high probability (IEEE Trial-Use Standard for Wireless Access in VehicularEnvironments (WAVE) - Multi-channel Operation (2006)). For a tutorial on WAVE protocol stack,we refer the reader to (Uzcategui & Acosta-Marum (2009)).Due to the relative novelty of DSRC and WAVE protocols, the majority of the widely usedVANET simulators do not implement the DSRC and WAVE protocols. One exception isthe NCTUNS simulation environment (Wang et al. (2003)), which implements both DSRC(IEEE 802.11p) and WAVE (IEEE 1609 set of standards) in its current version. Modeling thenetworking stack realistically is important for the credibility of the results obtained at eachlevel of the protocol stack, and especially for the application level, since all the potentialsimulation errors from the lower layers are reflected at the application layer. To this end, itwas recently shown that several stringent constraints exist in VANETs for applications (Bobanet al. (2009)), and even with the optimal settings with regards to the networking model, some


www.intechopen.com


Fig. 2. WAVE protocol stack.

of the results reported with simplified models, especially with regards to connectivity andmessage reachability (e.g., Palazzi et al. (2007)) are unachievable.

4. Signal propagation models

In order to adequately model the signal propagation in VANETs, appropriate models need tobe developed that take into account the unique characteristics of VANET environment (e.g.,high speed of the vehicles, obstruction-rich setting, specific location of the antennas, etc.).In the early days of VANET research, simple signal propagation models were utilized (e.g.,unit area disk model (Gupta & Kumar (2000)), free-space path loss (Goldsmith (2006)), amongothers), which were carried over from MANET research. Due to the significantly differentenvironment, these models do not provide satisfying accuracy for typical VANET scenarios(Koberstein et al. (2009)). Based on whether the model is accounting for a specific locationof the objects or generalized distribution of objects in the environment, we can distinguishdeterministic and stochastic models (Fig. 1). Deterministic models attempt to model thesignal behavior based on the exact environment in which the vehicle is currently located,and with specific locations of the objects surrounding the vehicle (Maurer et al. (2004)).Stochastic models, on the other hand, assume a location of the surrounding objects basedon a certain (often pre-defined) statistical distribution (Acosta & Ingram (2006)). Based onthe approach of modeling the environment, we distinguish geometrical or non-geometricalmodels. Geometrical models use the concepts of computational geometry to characterize theenvironment by generating the possible paths or rays between the transmitting and receivingvehicle. Non-geometrical models use the higher level properties of the environment (e.g.,path-loss exponent (Wang et al. (2004))) to approximate the signal power at the receiver.Furthermore, geometrical signal propagation models have to account for two types ofobstructions that affect the signal: static obstructions (e.g., road surface, buildings, overpasses,hills, etc.) and mobile obstructions (moving vehicles). Numerous studies have dealt withstatic obstacles as the key factors affecting signal propagation (Nagel & Eichler (2008) andGiordano et al. (2010)) and proposed models for accurately quantifying the impact of static


www.intechopen.com


obstacles. However, due to the nature of VANETs, where communication is often performedin V2V fashion, it is reasonable to expect that the moving vehicles will act as obstacles to thesignal, often affecting the signal propagation even more than static obstacles (e.g., in case ofan open road).Furthermore, the fact that the communicating entities in VANETs are vehicles exchangingdata in a V2V fashion raises new challenges in signal modeling. We observe, for example,that antenna heights of both transmitter and receiver are relatively low (on top of the vehiclesat best), such that other vehicles can act as obstacles for signal propagation by obstructingthe LOS between the communicating vehicles. The natural conclusion is that analyzingstatic obstacles only is not sufficient; vehicles as moving obstacles have to be taken intoaccount. These assumptions have been confirmed in several previous studied. Specifically,in (Otto et al. (2009)) V2V experiments were performed at 2.4 GHz frequency band in an openroad environment and pointed out a significantly worse signal reception on the same roadduring the traffic heavy, rush hour period in comparison to a no traffic, late night period.A similar experimental V2V study presented in (Takahashi et al. (2003)) analyzed the signalpropagation in “crowded” and “uncrowded” highway scenarios (depending on the numberof cars currently on the road) for the 60 GHz frequency band, and reported significantlyhigher path loss for the crowded scenarios. Several other studies (Jerbi et al. (2007), Wu et al.(2005), Matolak et al. (2005), and Vehicle Safety Communications Project, Final Report (2006)) hintthat other vehicles apart from the transmitter and receiver could be an important factor inmodeling the signal propagation by obstructing the LOS between the communicating vehicles.Despite this, virtually all of the state-of-the-art VANET simulators consider the vehicles asdimensionless entities that have no influence on signal propagation (Martinez et al. (2009)).This motivated our study on the impact of vehicles as obstacles on V2V communicationdescribed in (Boban et al. (2011)) and presented in the next section.

5. Model for Incorporating Vehicles as Obstacles in VANET Simulation

Environments1

5.1 Empirical measurements

In order to quantify the impact that the vehicles have on the received signal strength, weperformed experimental measurements. To isolate the effect of the obstructing vehicles, weaimed at setting up a controlled environment without other obstructions and with minimumimpact of other variables (e.g., other moving objects, electromagnetic radiation, etc). For thisreason, we performed experiments in an empty parking lot in Pittsburgh, PA (Fig. 3). Weanalyzed the received signal strength for the no obstruction, LOS case, and the non-LOScase where we introduced an obstructing vehicle (the van shown in Fig. 3) between thetransmitter (Tx) and the receiver (Rx) vehicles. The received signal strength was measuredfor the distances of 10, 50, and 100 m between the Tx and the Rx. In case of the non-LOSexperiments, the obstructing van was placed in the middle between the Tx and the Rx. Weperformed experiments at two frequency bands: 2.4 GHz (used by the majority of commercialWiFi devices) and 5.9 GHz (the band at which spectrum has been allocated for automotive useworldwide (IEEE Draft Standard IEEE P802.11p/D9.0 (July 2009))). For 2.4 GHz experiments,we equipped the Tx and Rx vehicles with laptops that had Atheros 802.11b/g wireless cardsinstalled and we used 3 dBi gain omnidirectional antennas.

1This section is based on the following paper: “Impact of Vehicles as Obstacles in Vehicular Ad HocNetworks”, IEEE Journal on Selected Area in Communications, c© [2011] IEEE: Boban et al. (2011)


www.intechopen.com


OBSTRUCTINGVEHICLETRANSMITTERRECEIVER TRANSMITTERRECEIVER

ANTENNA

Fig. 3. Experiment setup. c© [2011] IEEE

For 5.9 GHz experiments, we equipped the Tx and Rx vehicles with NEC Linkbird-MX devices(Festag et al. (2008)), which communicate via IEEE 802.11p wireless interfaces (IEEE DraftStandard IEEE P802.11p/D9.0 (July 2009)) and we used 5 dBi gain omnidirectional antennas.In both cases, antennas were mounted on the rooftops of the Tx and Rx vehicles (Fig. 3). Thedimensions of the vehicles are shown in Table 1, and the height of the antennas used in bothexperiments was 260 mm. The transmission power was set to 18 dBm. The Atheros wirelesscards in laptops as well as IEEE 802.11p radios in LinkBird-MXs were evaluated beforehandusing a real time spectrum analyzer and no significant power fluctuations were observed. Thecentral frequency was set to 2.412 GHz and 5.9 GHz, respectively, and the channel width was20 MHz. The data rate for 2.4 GHz experiments was 1 Mb/s, with 10 messages (140 bytesin size) sent per second using the ping command, whereas for 5.9 GHz experiments the datarate was 6 Mb/s (the lowest data rate in 802.11p for 20 MHz channel width) with 10 beacons(36 bytes in size) sent per second (Festag et al. (2008)). Each measurement was performedfor at least 120 seconds, thus resulting in a minimum of 1200 data packets transmittedper measurement. We collected the per-packet Received Signal Strength Indication (RSSI)information.Figures 4a and 4b show the RSSI for the LOS (no obstruction) and non-LOS (van obstructingthe LOS) measurements at 2.4 GHz and 5.9 GHz, respectively. The additional attenuationat both central frequencies ranges from approx. 20 dB at 10 m distance between Tx and Rxto 4 dB at 100 m. Even though the absolute values for the two frequencies differ (resultingmainly from the different quality radios used for 2.4 GHz and 5.9 GHz experiments), therelative trends indicate that the obstructing vehicles attenuate the signal more significantlythe closer the Tx and Rx are. To provide more insight into the distribution of the receivedsignal strength for LOS and non-LOS measurements, Fig. 5 shows the cumulative distributionfunction (CDF) of the RSSI measurements for 100 m in case of LOS and non-LOS at 2.4 GHz.The non-LOS case exhibits a larger variation and the two distributions are overall significantlydifferent, thus clearly showing the impact of the obstructing van. Similar distributions wereobserved for other distances between the Tx and the Rx.

Dimensions (m)Vehicle Height Width Length

2002 Lincoln LS (TX) 1.453 1.859 4.925

2009 Pontiac Vibe (RX) 1.547 1.763 4.371

2010 Ford E-250 (Obstruction) 2.085 2.029 5.504

Table 1. Dimensions of Vehicles. c© [2011] IEEE


www.intechopen.com


3540

LOS

1520253035

RSSI (dB)

LOS

Obstructed

05

1015

10m 50m 100mDistance

(a) 2.4 GHz

3540

LOS

1520253035

RSSI (dB)

LOS

Obstructed

05

1015

10m 50m 100mDistance

(b) 5.9 GHz

Fig. 4. RSSI measurements: average RSSI with and without the obstructing vehicle.c© [2011] IEEE

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

CDF

RSSI (dB)

100 m - No Obstruction

100 m - Obstructing Van

Fig. 5. Distribution of the RSSI for 100 m in case of LOS (no obstruction) and non-LOS(obstructing van) at 2.4 GHz. c© [2011] IEEE

5.2 Model analysis

5.2.1 The impact of vehicles on line of sight

In order to isolate and quantify the effect of vehicles as obstacles on signal propagation, we donot consider the effect of other obstacles such as buildings, overpasses, vegetation, or otherroadside objects on the analyzed highways. Since those obstacles can only further reduce theprobability of LOS, our approach leads to a best case analysis for probability of LOS.Figure 6 describes the methodology we use to quantify the impact of vehicles as obstacleson LOS in a V2V environment. Using aerial imagery (Fig. 6a) to obtain the location andlength of vehicles, we devise a model that is able to analyze all possible connections betweenvehicles within a given range (Fig. 6b). For each link – such as the one between the vehiclesdesignated as transmitter (Tx) and receiver (Rx) in Fig. 6b – the model determines the existenceor non-existence of the LOS based on the number and dimensions of vehicles potentiallyobstructing the LOS (in case of the aforementioned vehicles designated as Tx and Rx, thevehicles potentially obstructing the LOS are those designated as Obstacle 1 and Obstacle 2 inFig. 6b).The proposed model calculates the (non-)existence of the LOS for each link (i.e., between allcommunicating pairs) in a deterministic fashion, based on the dimensions of the vehiclesand their locations. However, in order to make the model mathematically tractable, we


www.intechopen.com


Tx

Rx

Obstacle 1

Obstacle 2

(a) Aerial photography (b) Abstracted model showing possible

connections

LOS not obstructed

LOS potentially obstructed

60% of First

Fresnel Ellipsoid

Tx RxObstacle 1 Obstacle 2d

dobs1

dobs2

h1h i h jh2

(c) P(LOS) calculation for a given link

Fig. 6. Model for evaluating the impact of vehicles as obstacles on LOS (for simplicity, vehicleantenna heights (ha) are not shown in subfigure (c)). c© [2011] IEEE


www.intechopen.com


derive the expressions for the microscopic (i.e., per-link and per-node) and macroscopic (i.e.,system-wide) probability of LOS. It has to be noted that, from the electromagnetic wavepropagation perspective, the LOS is not guaranteed with the existence of the visual sight linebetween the Tx and Rx. It is also required that the Fresnel ellipsoid is free of obstructions((Rappaport, 1996, Chap. 3)). Any obstacle that obstructs the Fresnel ellipsoid might affectthe transmitted signal. As the distance between the transmitter and receiver increases, thediameter of the Fresnel ellipsoid increases accordingly. Besides the distance between the Txand Rx, the Fresnel ellipsoid diameter is also a function of the wavelength.As we will show later in Section 5.3, the vehicle heights follow a normal distribution. Tocalculate P(LOS)ij, i.e., the probability of LOS for the link between vehicles i and j, with onevehicle as a potential obstacle between Tx and Rx (of height hi and hj, respectively), we have:

P(LOS|hi, hj) = 1 − Q

(

h − µ

σ

)

(1)

and

h = (hj − hi)dobs

d+ hi − 0.6r f + ha, (2)

where the i, j subscripts are dropped for clarity, and h denotes the effective height of thestraight line that connects Tx and Rx at the obstacle location when we consider the first Fresnelellipsoid. Furthermore, Q(·) represents the Q-function, µ is the mean height of the obstacle,σ is the standard deviation of the obstacle’s height, d is the distance between the transmitterand receiver, dobs is the distance between the transmitter and the obstacle, ha is the height ofthe antenna, and r f is the radius of the first Fresnel zone ellipsoid which is given by

r f =

√

λdobs(d − dobs)

d,

with λ denoting the wavelength. We use the appropriate λ for the proposed standard forVANET communication (DSRC), which operates in the 5.9 GHz frequency band. In ourstudies, we assume that the antennas are located on top of the vehicles in the middle of the roof(which was experimentally shown to be the overall optimum placement of the antenna (Kaulet al. (2007))), and we set the ha to 10 cm. As a general rule commonly used in literature, LOSis considered to be unobstructed if intermediate vehicles obstruct the first Fresnel ellipsoid byless than 40% ((Rappaport, 1996, Chap. 3)). Furthermore, for No vehicles as potential obstaclesbetween the Tx and Rx, we get (see Fig. 6c)

P(LOS|hi, hj) =No

∏k=1

[

1 − Q

(

hk − µk

σk

)]

, (3)

where hk is the effective height of the straight line that connects Tx and Rx at the location of thek-th obstacle considering the first Fresnel ellipsoid, µk is the mean height of the k-th obstacle,and σk is the standard deviation of the height of the k-th obstacle.Averaging over the transmitter and receiver antenna heights with respect to the road, weobtain the unconditional P(LOS)ij

P(LOS)ij =∫ ∫

P(LOS|hi, hj)p(hi)p(hj)dhidhj, (4)

where p(hi) and p(hj) are the probability density functions for the transmitter and receiverantenna heights with respect to the road, respectively.


www.intechopen.com


The average probability of LOS for a given vehicle i, P(LOS)i, and all its Ni neighbors isdefined as

P(LOS)i =1

Ni

Ni

∑j=1

P(LOS)ij (5)

To determine the system-wide ratio of LOS paths blocked by other vehicles, we averageP(LOS)i over all Nv vehicles in the system, yielding

P(LOS) =1

Nv

Nv

∑i=1

P(LOS)i. (6)

Furthermore, we analyze the behavior of the probability of LOS for a given vehicle i over time.Let us denote the i-th vehicle probability of LOS at a given time t as P(LOS)t

i . We define thechange in the probability of LOS for the i-th vehicle over two snapshots at times t1 and t2 as

∆P(LOS)i = |P(LOS)t2

i − P(LOS)t1

i |, (7)

where P(LOS)t1

i and P(LOS)t2

i are obtained using (5).It is important to note that equations (1) to (7) depend on the distance between the node iand the node j (i.e., transmitter and receiver) in a deterministic manner. More specifically, thesnapshot obtained from aerial photography provides the exact distance d (Fig. 6c) betweenthe nodes i and j. While in our study we used aerial photography to get this information, anyVANET simulator would also provide the exact location of vehicles based on the assumedmobility model (e.g., car-following (Rothery (1992)), cellular automata (Tonguz et al. (2009)),etc.), hence the distance d between the nodes i and j would still be available. This alsoexplains why the proposed model is independent of the simulator used, since it can beincorporated into any VANET simulator, regardless of the underlying mobility model, aslong as the locations of the vehicles are available. Furthermore, even though we used thehighway environment for testing, the proposed model can be used for evaluating the impactof obstructing vehicles on any type of road, irrespective of the shape of the road (e.g., singleor multiple lanes, straight or curvy) or location (e.g., highway, suburban, or urban2).

5.2.2 The impact of vehicles on signal propagation

The attenuation on a radio link increases if one or more vehicles intersect the ellipsoidcorresponding to 60% of the radius of the first Fresnel zone, independent of their positionson the Tx-Rx link (Fig. 6c). This increase in attenuation is due to the diffraction of theelectromagnetic waves. The additional attenuation due to diffraction depends on a variety offactors: the obstruction level, the carrier frequency, the electrical characteristics, the shape ofthe obstacles, and the amount of obstructions in the path between transmitter and receiver. Tomodel vehicles obstructing the LOS, we use the knife-edge attenuation model. It is reasonableto expect that more than one vehicle can be located between transmitter (Tx) and receiver (Rx).Thus, we employ the multiple knife-edge model described in ITU-R recommendation (ITU-R(2007)). When there are no vehicles obstructing the LOS between the Tx and Rx, we use the

2However, to precisely quantify the impact of obstructing vehicles in complex urban environments,further research is needed to determine the interplay between the vehicle-induced obstruction and theobstruction caused by other objects (e.g., buildings, overpasses, etc.).


www.intechopen.com


free space path loss model (Goldsmith (2006))3.Single Knife-EdgeThe simplest obstacle model is the knife-edge model, which is a reference case for morecomplex obstacle models (e.g., cylinder and convex obstacles). Since the frequency of DSRCradios is 5.9 GHz, the knife-edge model theoretically presents an adequate approximation forthe obstacles at hand (vehicles), as the prerequisite for the applicability of the model, namelya significantly smaller wavelength than the size of the obstacles (ITU-R (2007)), is fulfilled (thewavelength of the DSRC is approximately 5 cm, which is significantly smaller than the size ofthe vehicles).The obstacle is seen as a semi-infinite perfectly absorbing plane that is placed perpendicularto the radio link between the Tx and Rx. Based on the Huygens principle, the electric field isthe sum of Huygens sources located in the plane above the obstruction and can be computedby solving the Fresnel integrals (Parsons (2000)). A good approximation for the additionalattenuation (in dB) due to a single knife-edge obstacle Ask can be obtained using the followingequation (ITU-R (2007)):

Ask =

⎧

⎪

⎨

⎪

⎩

6.9 + 20log10

[

√

(v − 0.1)2 + 1 + v − 0.1]

;

for v > −0.70; otherwise,

(8)

where v =√

2H/r f , H is the difference between the height of the obstacle and the height ofthe straight line that connects Tx and Rx, and r f is the Fresnel ellipsoid radius.Multiple Knife-EdgeThe extension of the single knife-edge obstacle case to the multiple knife-edge is notimmediate. All of the existing methods in the literature are empirical and the results varyfrom optimistic to pessimistic approximations (Parsons (2000)). The method in (Epstein &Peterson (1953)) presents a more optimistic view, whereas the methods in (Deygout (1966))and (Giovaneli (1984)) are more pessimistic approximations of the real world. Usually,the pessimistic methods are employed when it is desirable to guarantee that the systemwill be functional with very high probability. On the other hand, the more optimisticmethods are used when analyzing the effect of interfering sources in the communicationsbetween transmitter and receiver. To calculate the additional attenuation due to vehicles, weemploy the ITU-R method (ITU-R (2007)), which can be seen as a modified version of theEpstein-Patterson method, where correcting factors are added to the attenuation in order tobetter approximate reality.

5.3 Model requirements

The model proposed in the previous section is aimed at evaluating the impact of vehicles asobstacles using geometry concepts and relies heavily on realistic modeling of the physicalenvironment. In order to employ the proposed model accurately, realistic modeling of thefollowing physical properties is necessary: determining the exact position of vehicles andthe inter-vehicle spacing; determining the speed of vehicles; and determining the vehicledimensions.

3We acknowledge that the free space model might not be the best approximation of the LOScommunication on the road. However, due to its tractability, it allows us to analyze the relationshipbetween the LOS and non-LOS conditions in a deterministic manner.


www.intechopen.com


Dataset Size # vehicles # large vehicles Veh. density

A28 12.5 km 404 58 (14.36%) 32.3 veh/km

A3 7.5 km 55 10 (18.18%) 7.3 veh/km

Table 2. Analyzed highway datasets. c© [2011] IEEE

Determining the exact position of vehicles and the inter-vehicle spacing

The position and the speed of vehicles can easily be obtained from any currently availableVANET mobility model. However, in order to test our methodology with the most realisticparameters available, we used aerial photography. This technique is used by the trafficengineering community as an alternative to ground-based traffic monitoring (McCasland, WT (1965)), and was recently applied to VANET connectivity analysis (Ferreira et al. (2009)). It iswell suited to characterize the physical interdependencies of signal propagation and vehiclelocation, because it gives the exact position of each vehicle. We analyzed two distinct datasets, namely two Portuguese highways near the city of Porto, A28 and A3, both with fourlanes (two per direction). Detailed parameters for the two datasets are presented in Table 2.For an extensive description of the method used for data collection and analysis, we refer thereader to (Ferreira et al. (2009)).

Determining the speed of vehicles

For the observed datasets, besides the exact location of vehicles and the inter-vehicle distances,stereoscopic imagery was once again used to determine the speed and heading of vehicles.Since the successive photographs were taken with a fixed time interval (5 seconds), bymarking the vehicles on successive photographs we were able to measure the distance thevehicle traversed, and thus infer the speed and heading of the vehicle. The measured speedand inter-vehicle spacing is used to analyze the behavior of vehicles as obstacles while theyare moving.The distribution of inter-vehicle spacing for both cases can be well fitted with an exponentialprobability distribution. This agrees with the empirical measurements made on the I-80interstate in California reported in (Wisitpongphan et al. (Oct. 2007)). The speed distributionon both highways is well approximated by a normal probability distribution. Table 3 showsthe parameters of best fits for inter-vehicle distances and speeds.

Determining the vehicle dimensions

From the photographs, we were also able to obtain the length of each vehicle accurately,however the width and height could not be determined with satisfactory accuracy due toresolution constraints and vehicle mobility. To assign proper widths and heights to vehicles,we use the data made available by the Automotive Association of Portugal (AssociacaoAutomovel de Portugal (n.d.)), which issued an official report about all vehicles currently incirculation in Portugal. From the report we extracted the eighteen most popular personalvehicle brands which comprise 92% of all personal vehicles circulating on Portuguese roads,and consulted an online database of vehicle dimensions (Automotive Technical Data andSpecifications (n.d.)) to arrive at the distribution of height and width required for our analysis.The dimensions of the most popular personal vehicles showed that both the vehicle widthsand heights can be modeled as a normal random variable. Detailed parameters for thefitting process for both personal and large vehicles are presented in Table 4. For both widthand height of personal vehicles, the standard error for the fitting process remained below0.33% for both the mean and the standard deviation. The data regarding the specific types


www.intechopen.com


Data for A28

Parameter Estimate Std. Error

Speed: normal fit

mean (km/h) 106.98 1.05

std. deviation (km/h) 21.09 0.74

Inter-vehicle spacing : exponential fit

mean (m) 51.58 2.57

Data for A3

Parameter Estimate Std. Error

Speed: normal fit

mean (km/h) 122.11 3.97

std. deviation (km/h) 28.95 2.85

Inter-vehicle spacing : exponential fit

mean (m) 215.78 29.92

Table 3. Parameters of the Best Fit Distributions for Vehicle Speed and Inter-vehicle spacing.c© [2011] IEEE

of large vehicles (e.g., trucks, vans, or buses) currently in circulation was not available.Consequently, the precise dimension distributions of the most representative models couldnot be obtained. For this reason, we infer large vehicle height and width values from the dataavailable on manufacturers’ websites, which can serve as rough dimension guidelines thatshow significantly different height and width in comparison to personal vehicles.

5.4 Computational complexity of the proposed model

In order to determine the LOS conditions between two neighboring nodes, we analyzed theexistence of LOS in a three dimensional space, as shown in Fig. 6 and explained in the previoussections. Our model for determining the existence of LOS between vehicles and, in caseof obstruction, obtaining the number and location of the obstructions, belongs to a class ofcomputational geometry problems known as geometric intersection problems (de Berg et al.(1997)), which deal with pairwise intersections between line segments in an n-dimensionalspace. These problems occur in various contexts, such as computer graphics (object occlusion)and circuit design (crossing conductors), amongst others (Bentley & Ottmann (1979)).Specifically, for a given number of line segments N, we are interested in determining,reporting, and counting the pairwise intersections between the line segments. For our specificapplication, the line segments of interest are of two kinds: a) the LOS rays between thecommunicating vehicles (lines colored red in Fig. 6b); and b) the lines that compose thebounding rectangle representing the vehicles (lines colored blue in Fig. 6b). It has to benoted that the intersections of interest are only those between the LOS rays and the boundingrectangle lines, and not between the lines of the same type. Therefore, we arrive at aspecial case of the segment intersection problem, namely the so-called “red-blue” intersectionproblem. Given a set of red line segments r and a set of blue line segments b, with a total ofN = r+ b segments, the goal is to report all K intersections between red and blue segments, forwhich an efficient algorithm was presented in (Agarwal (1991)). The time-complexity of thealgorithm proposed in (Agarwal (1991)), using the randomized approach of (Clarkson (1987)),is O(N4/3 log N + K), where K is the number of red-blue intersections, with space complexityof O(N4/3). This algorithm fits our purposes perfectly, as the red segments correspond to


www.intechopen.com


Personal vehicles

Parameter Estimate

Width: normal fit

mean (cm) 175

std. deviation (cm) 8.3

Height: normal fit

mean (cm) 150


Large vehicles

Parameter Estimate

Width: constant

mean (cm) 250

Height: normal fit

mean (cm) 335


Table 4. Parameters of the Best Fit Distributions for Vehicle Width and Height. c© [2011] IEEE

the LOS rays between the communicating vehicles and blue segments are the lines of thebounding rectangles representing the vehicles (see Fig. 6b).To assign physical values to r and b, we denote v as the number of vehicles in the systemand v′ as the number of transmitting vehicles. The number of LOS rays results in r = Cv′,where the average number of neighbors C is an increasing function of the vehicle densityand transmission range. The number of lines composing the bounding vehicle rectangles canbe expressed as b = 4v, since each vehicle is represented by four lines forming a rectangle(see Fig. 6b). Therefore, a more specific time-complexity bound can be written as O((Cv′ +4v)4/3 log(Cv′ + 4v) + K).Apart from the algorithm for determining the red-blue intersections, the rest of the proposedmodel consists in calculating the additional signal attenuation due to vehicles for eachcommunicating pair. In the case of non-obstructed LOS the algorithm terminates, whereas forobstructed LOS, the red-blue intersection algorithm is used to store the number and locationof intersecting blue lines (representing obstacles). The total number of intersections is givenby K = gr, where g is the number of obstacles (i.e., vehicles) in the LOS path and is a subsetof C. The complete algorithm for additional attenuation due to vehicles is implemented asfollows.The function getIntersect(·) is based on the aforementioned red-blue line intersectionalgorithm (Agarwal (1991)), and has complexity O((Cv′ + 4v)4/3 log(Cv′ + 4v)+ gr), whereasthe function calcAddAtten(·) is based on multiple knife-edge attenuation model describedin (ITU-R (2007)) with time-complexity of O(g2) for each LOS ray r. It follows that thetime-complexity of the entire algorithm is given by O((Cv′ + 4v)4/3 log(Cv′ + 4v) + g2r).In order to implement the aforementioned algorithm in VANET simulators, apart fromthe information available in the current VANET simulators, very few additional pieces ofinformation are necessary. Specifically, the required information pertains to the physicaldimensions of the vehicles. Apart from this, the model only requires the information on theposition of the vehicles at each simulation time step. This information is available in anyvehicular mobility model currently in use in VANET simulators.


www.intechopen.com


Algorithm 1 Calculate additional attenuation due to vehicles

for i = 1 to r do[coord] = getIntersect(i) {For each LOS ray in r, obtain the location of intersections as per(Agarwal (1991))}if size([coord]) �= 0 then

att = calcAddAtten([coord]) {Calculate the additional attenuation due to vehicles asper (ITU-R (2007))}

elseatt = 0 dB {Additional attenuation due to vehicles equals zero.}

end ifend for

Highways

Transmission Range (m)Highway 100 250 500

A3 P(LOS) 0.8445 0.6839 0.6597

A28 P(LOS) 0.8213 0.6605 0.6149

Table 5. P(LOS) for A3 and A28. c© [2011] IEEE

5.5 Results

We implemented the model described in previous sections in Matlab. In this section wepresent the results based on testing the model using the A3 and A28 datasets. We alsopresent the results of the empirical measurements that we performed in order to characterizethe impact of the obstructing vehicles on the received signal strength. We emphasize thatthe model developed in the paper is not dependent on these datasets, but can be usedin any environment by applying the analysis presented in Section 5.2. Furthermore, theobservations pertaining to the inter-vehicle and speed distributions on A3 and A28 are usedonly to characterize the behavior of the highway environment over time. We do not use thesedistributions in our model; rather, we use actual positions of the vehicles. Since the modeldeveloped in Section 5.2 is intended to be utilized by VANET simulators, the positions of thevehicles can easily be obtained through the employed vehicular mobility model.We first give evidence that vehicles as obstacles have a significant impact on LOScommunication in both sparse (A3) and more dense (A28) networks. Next, we analyze themicroscopic probability of LOS to determine the variation of the LOS conditions over time fora given vehicle. Then, we used the speed and heading information to characterize both themicroscopic and macroscopic behavior of the probability of LOS on highways over time inorder to determine how often the proposed model needs to be recalculated in the simulators,and to infer the stationarity of the system-wide probability of LOS. Using the employedmultiple knife-edge model, we present the results pertaining to the decrease of the receivedpower and packet loss for DSRC due to vehicles. Finally, we corroborate our findings on theimpact of the obstructing vehicles and discuss the appropriateness of the knife-edge modelby performing empirical measurements of the received signal strength in LOS and non-LOSconditions.


www.intechopen.com


100 300 500 7500

10

20

30

40

50

Transmission range (meters)

Avera

ge N

um

ber

of

Neig

hbors

LOS Not Obstructed

LOS Obstructed

Fig. 7. Average number of neighbors with unobstructed and obstructed LOS on A28highway. c© [2011] IEEE

5.5.1 Probability of line of sight

Macroscopic probability of line of sight. Table 5 presents the values of P(LOS) with respect to theobserved range on highways. The highway results show that even for the sparsely populated

A3 highway the impact of vehicles on P(LOS) is significant. This can be explained by theexponential inter-vehicle spacing, which makes it more probable that the vehicles are locatedclose to each other, thus increasing the probability of having an obstructed link between twovehicles. For both highways, it is clear that the impact of other vehicles as obstacles can notbe neglected even for vehicles that are relatively close to each other (for the observed range of

100 m, P(LOS) is under 85% for both highways, which means that there is a non-negligible15% probability that the vehicles will not have LOS while communicating). To confirm theseresults, Fig. 7 shows the average number of neighbors with obstructed and unobstructed LOSfor the A28 highway. The increase of obstructed vehicles in both absolute and relative senseis evident.Microscopic probability of line of sight. In order to analyze the variation of the probability of LOSfor a vehicle and its neighbors over time, we observe the ∆P(LOS)i (as defined in equation(7)) on A28 highway for the maximum communication range of 750 m. Table 6 shows the∆P(LOS)i. The variation of probability of LOS is moderate for periods of seconds (even for thelargest offset of 2 seconds, only 15% of the nodes have the ∆P(LOS)i greater than 20%). Thisresult suggests that the LOS conditions between a vehicle and its neighbors will remain largelyunchanged for a period of seconds. Therefore, a simulation time-step of the order of secondscan be used for calculations of the impact of vehicles as obstacles. From a simulation executionstandpoint, the time-step of the order of seconds is quite a long time when compared withthe rate of message transmission, measured in milliseconds; this enables a more efficient andscalable design and modeling of vehicles as obstacles on a microscopic, per-vehicle level. Withthe proper implementation of the LOS intersection model discussed in Sections 5.2 and 5.4, the modeling of vehicles as obstacles should not induce a large overhead in the simulationexecution time.

5.5.2 Received power

Based on the methodology developed in Section 5.2, we utilize the multiple knife-edge modelto calculate the additional attenuation due to vehicles. We use the obtained attenuation tocalculate the received signal power for the DSRC. We employed the knife-edge model for its


www.intechopen.com


∆P(LOS)i in %Time offset < 5% 5-10% 10-20% >20%

1ms 100% 0% 0% 0%

10ms 99% 1% 0% 0%

100ms 82% 15% 3% 0%

1s 35% 33% 22% 10%

2s 31% 25% 29% 15%

Table 6. Variation of P(LOS)i over time for the observed range of 750 m on A28.c© [2011] IEEE

−110 −100 −90 −80 −70 −60 −50 −400

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Received Power Level (dBm)

Cum

ula

tive D

istr

ibution F

unction

Rx sensitivity level

Vehicles as obstacles

Free Space

Fig. 8. The impact of vehicles as obstacles on the received signal power on highway A28.

simplicity and the fact that it is well studied and often used in the literature. However, wepoint out that the LOS analysis and the methodology developed in Section 5.2 can be used inconjunction with any channel model that relies on the distinction between the LOS and NLOScommunication (Zang et al. (2005) and Wang et al. (2004)).For the A28 highway and the observed range of 750 m, with the transmit power set to 18 dBm,3 dBi antenna gain for both transmitters and receivers, at the 5.9 GHz frequency band, theresults for the free space path loss model (Goldsmith (2006)) (i.e., not including vehicles asobstacles) and our model that accounts for vehicles as obstacles are shown in Fig. 8. Theaverage additional attenuation due to vehicles was 9.2 dB for the observed highway.Using the minimum sensitivity thresholds as defined in the DSRC standard (see Table 7)(Standard Specification for Telecommunications and Information Exchange Between Roadside andVehicle Systems - 5GHz Band Dedicated Short Range Communications (DSRC) Medium Access

Data Rate (Mb/s) Modulation Minimum sensitivity (dBm)

3 BPSK −85

4.5 BPSK −84

6 QPSK −82

9 QPSK −80

12 QAM-16 −77

18 QAM-16 −70

24 QAM-64 −69

27 QAM-64 −67

Table 7. Requirements for DSRC Receiver Performance. c© [2011] IEEE


www.intechopen.com


100 250 375 500 625 7500.4

0.5

0.6

0.7

0.8

0.9

1

Transmission Range (m)P

acket

sucess r

ate

3Mb/s − obstacles

6Mb/s − obstacles

12Mb/s − obstacles

3Mb/s − free space



Fig. 9. The impact of vehicles as obstacles on packet success rate for various DSRC data rateson A28 highway. c© [2011] IEEE

Control (MAC) and Physical Layer (PHY) Specifications (Sep. 2003)), we calculate the packetsuccess rate (PSR, defined as the ratio of received messages to sent messages) as follows. Weanalyze all of the communicating pairs within an observed range, and calculate the receivedsignal power for each message. Based on the sensitivity thresholds presented in Table 7, wedetermine whether a message is successfully received. For the A28 highway, Fig. 9 shows thePSR difference between the free space path loss and the implemented model with vehicles asobstacles for rates of 3, 6, and 12 Mb/s. The results show that the difference is significant, asthe percentage of lost packets can be up to 25% higher when vehicles are accounted for.These results show that not only do the vehicles significantly decrease the received signalpower, but the resulting received power is highly variable even for relatively short distancesbetween the communicating vehicles, thus calling for a microscopic, per-vehicle analysis ofthe impact of obstructing vehicles. Models that try to average the additional attenuation dueto vehicles could fail to describe the complexity of the environment, thus yielding unrealisticresults. Furthermore, the results show that the distance itself can not be solely used fordetermining the received power, since even the vehicles close by can have a number of othervehicles obstructing the communication path and therefore the received signal power becomesworse than for vehicles further apart that do not have obstructing vehicles between them.

5.6 Discussion

In this section, we describe the impact that the obtained results have on various aspects ofV2V communication modeling, ranging from physical to application layer to the realism ofVANET simulators.

5.6.1 Impact on signal propagation modeling

The results presented in this paper clearly indicate that vehicles as obstacles have a significantimpact on signal propagation; therefore, in order to properly model V2V communication, it isimperative that vehicles as obstacles are accounted for. Furthermore, the effect of vehiclesas obstacles cannot be neglected even in the case of relatively sparse vehicular networks,as one of the two analyzed highway datasets showed (namely, the dataset collected on theA3 highway). Therefore, previous efforts pertaining to signal propagation modeling in V2Vcommunication which do not account for vehicles as obstacles, can be deemed as optimisticin overestimating the received signal power level.


www.intechopen.com


5.6.2 Impact on data link layer

Neglecting vehicles as obstacles on the physical layer has profound effects on the performanceof upper layers of the communication stack. The effects on the data link layer are twofold: a)the medium contention is overestimated in models that do not include vehicles as obstacles inthe calculation, thus potentially representing a more pessimistic situation than the real-worldwith regards to contention and collision; and b) the network reachability is bound to beoverestimated, due to the fact that the signal is considered to reach more neighbors and at ahigher power than in the real world. These results have important implications for vehicularMedium Access Control (MAC) protocol design; MAC protocols will have to cope with anincreased number of hidden vehicles due to other vehicles obstructing them.

5.6.3 Impact on the design of routing protocols

If vehicles as obstacles are not accounted for, the impact on routing protocols is representedby an overly optimistic hop count; in the process of routing, next hop neighbors are selectedthat are actually not within the reach of the current transmitter, thus inducing an unrealisticbehavior of the routing protocol, as the message is considered to reach the destination with asmaller number of hops than it is actually required.As an especially important class of routing protocols, safety messaging protocols, are oftenmodeled and evaluated using distance information only. As our results have shown, notaccounting for vehicles as obstacles in such calculations results in the overestimation of thenumber of reachable neighbors, which yields unrealistic results with regards to networkreachability and message penetration rate. Therefore, it is extremely important to accountfor vehicles as obstacles in V2V, especially since safety applications running over suchprotocols require that practically all vehicles receive the message, thus posing very stringentrequirements on the routing protocols.For these reasons, it is more beneficial to design routing protocols that rely primarily on thereceived signal strength instead of the geographical location of vehicles, since this wouldensure that the designated recipient is actually able to receive the message. However, evenwith smart protocols that are able to properly evaluate the channel characteristics betweenthe vehicles, in case of lower market penetration rates of the communicating equipment,the vehicles that are not equipped could significantly hinder the communication betweenthe equipped vehicles; this is another aspect of routing protocol design that is significantlyaffected by the impact of vehicles as obstacles in V2V communication.Similarly, the results suggest that, where available, vehicle-to-infrastructure (V2I)communication (where vehicles are communicating with road side equipment) should befavored instead of V2V communication; since the road side equipment is supposed to beplaced in lamp posts, traffic lights, or on the gantries above the highways such as the one inthe Fig. 6a), all of which are located 3-6 meters above ground level, other vehicles as obstacleswould impact the LOS much less than in the case of V2V communication. Therefore, similarlyto differentiating vehicles with regards to their dimensions, routing protocols would benefitfrom being able to differentiate between the road side equipment and vehicles.

5.6.4 Impact on VANET simulations

VANET simulation environments have largely neglected the modeling of vehicles as obstaclesin V2V communication. Results presented in this paper showed that the vehicles have asignificant impact on the LOS, and in order to realistically model the V2V communicationin simulation environments, vehicles as obstacles have to be accounted for. This implies that


www.intechopen.com


the models that relied on the simulation results that did not account for vehicles as obstacleshave been at best producing an optimistic upper bound of the results that can be expected inthe real world.In order to improve the realism of the simulators and to enable the implementationof a scalable and realistic framework for describing the vehicles as obstacles in V2Vcommunication, we proposed a simple yet realistic model for determining the probability ofLOS on both macroscopic and microscopic level. Using the results that proved the stationarityof the probability of LOS, we showed that the average probability of LOS does not changeover time if the vehicle arrival rate remains constant. Furthermore, over a period of seconds,the LOS conditions remain mostly constant even for the microscopic, per-vehicle case. Thisimplies that the modeling of the impact of vehicles as obstacles can be performed at the rateof seconds, which is two to three orders of magnitude less frequent than the rate of messageexchange (most often, messages are exchanged on a millisecond basis). Therefore, with theproper implementation of the proposed model, the calculation of the impact of vehicles onLOS should not induce a large overhead in the simulation execution time.

6. Conclusions

We discussed the state-of-the-art in VANET modeling and simulation, and described thebuilding blocks of VANET simulation environments, namely the mobility, networkingand signal propagation models. We described the most important models for each ofthese categories, and we emphasized that several areas are not optimally representedin state-of-the-art VANET simulators. Namely, the vehicle interaction and traffic ruleenforcement models in most current simulators leave a lot to be desired, and the lack of WAVEand DSRC protocol implementation in the simulators is also a fact for most simulators. Finally,we pointed out that the models for moving obstacles are lacking in modern simulators, andwe described our proposed model for vehicles as physical obstacles in VANETs as follows.First, using the experimental data collected in a measurement campaign, and by utilizingthe real world data collected by means of stereoscopic aerial photography, we showed thatvehicles as obstacles have a significant impact on signal propagation in V2V communication;in order to realistically model the communication, it is imperative that vehicles as obstaclesare accounted for. The obtained results point out that vehicles are an important factor in bothhighway and urban, as well as in sparse and dense networks. Next, we characterized thevehicles as three-dimensional objects that can obstruct the LOS between the communicatingpair. Then, we modeled the vehicles as physical obstacles that attenuate the signal, whichallowed us to determine their impact on the received signal power, and consequently on thepacket error rate. The presented model is computationally efficient and, as the results showed,can be updated at a rate much lower than the message exchange rate in VANETs. Therefore,it can easily be implemented in any VANET simulation environment to increase the realism.

7. References

Acosta, G. & Ingram, M. (2006). Model development for the wideband expresswayvehicle-to-vehicle 2.4 GHz channel, IEEE Wireless Communications and NetworkingConference, 2006. WCNC 2006., Vol. 3, pp. 1283–1288.

Agarwal, P. K. (1991). Intersection and Decomposition Algorithms for Planar Arrangements,Cambridge University Press.

Associacao Automovel de Portugal (n.d.).


www.intechopen.com


URL: http://www.acap.pt/Automotive Technical Data and Specifications (n.d.).

URL: http://www.carfolio.com/Bai, F., Elbatt, T., Hollan, G., Krishnan, H. & Sadekar, V. (2006). Towards characterizing

and classifying communication-based automotive applications from a wirelessnetworking perspective, 1st IEEE Workshop on Automotive Networking and Applications(AutoNet) .

Bai, F., Sadagopan, N. & Helmy, A. (2003). IMPORTANT: a framework to systematicallyanalyze the impact of mobility on performance of routing protocols for adhocnetworks, INFOCOM 2003. Twenty-Second Annual Joint Conference of the IEEEComputer and Communications Societies. IEEE 2: 825–835 vol.2.

Bentley, J. & Ottmann, T. (1979). Algorithms for reporting and counting geometricintersections, Computers, IEEE Transactions on C-28(9): 643–647.

Boban, M., Tonguz, O. & Barros, J. (2009). Unicast communication in vehicular ad hocnetworks: a reality check, IEEE Communications Letters 13(12): 995–997.

Boban, M., Vinhoza, T. T. V., Barros, J., Ferreira, M. & Tonguz, O. K. (2011). Impact ofVehicles as Obstacles in Vehicular Ad Hoc Networks, IEEE Journal on Selected Areas inCommunications 29(1): 15–28.

Cheng, Y. & Robertazzi, T. (Jul. 1989). Critical Connectivity Phenomena in Multihop RadioModels, IEEE Transactions on Communications 37(7): 770–777.

Choffnes, D. R. & Bustamante, F. E. (2005). An integrated mobility and traffic model forvehicular wireless networks, VANET ’05: Proceedings of the 2nd ACM internationalworkshop on Vehicular ad hoc networks, ACM, New York, NY, USA, pp. 69–78.

Clarkson, K. (1987). New applications of random sampling in computational geometry,Discrete and Computational Geometry 2: 195–222.

CORSIM: Microscopic Traffic Simulation Model (n.d.).URL: http://www-mctrans.ce.ufl.edu/featured/TSIS/

Davies, J. J., Beresford, A. R. & Hopper, A. (2006). Scalable, distributed, real-time mapgeneration, IEEE Pervasive Computing 5(4): 47–54.

de Berg, M., van Kreveld, M., Overmars, M. & Schwarzkopf, O. (1997). Computational GeometryAlgorithms and Applications, Springer-Verlag.

Deygout, J. (1966). Multiple knife-edge diffraction of microwaves, IEEE Transactions onAntennas and Propagation 14(4): 480–489.

Epstein, J. & Peterson, D. W. (1953). An experimental study of wave propagatin at 850MC,Proceedings of the IRE 41(5): 595–611.

Ferreira, M., Conceicao, H., Fernandes, R. & Tonguz, O. K. (2009). StereoscopicAerial Photography: An Alternative to Model-Based Urban Mobility Approaches,Proceedings of the Sixth ACM International Workshop on VehiculAr Inter-NETworking(VANET 2009), ACM New York, NY, USA.

Festag, A., Baldessari, R., Zhang, W., Le, L., Sarma, A. & Fukukawa, M. (2008). Car-2-xcommunication for safety and infotainment in europe, NEC Technical Journal 3(1).

Giordano, E., Frank, R., Pau, G. & Gerla, M. (2010). Corner: a realistic urban propagationmodel for vanet, WONS’10: Proceedings of the 7th international conference on Wirelesson-demand network systems and services, IEEE Press, Piscataway, NJ, USA, pp. 57–60.

Giovaneli, C. L. (1984). An analysis of simplified solutions for multiple knife-edge diffraction,IEEE Transactions on Antennas and Propagation 32(3): 297–301.

Gipps, P. G. (1986). A model for the structure of lane-changing decisions, Transportation


www.intechopen.com


Research Part B: Methodological 20(5): 403–414.Goldsmith, A. J. (2006). Wireless Communications, Cambridge University Press.Gupta, P. & Kumar, P. (2000). The Capacity of Wireless Networks, IEEE Transactions on

Information Theory 46(2): 388–404.Harri, J. (2010). Vehicular mobility modeling for vanet, VANET Vehicular Applications and

Inter-Networking Technologies, Wiley, pp. 107–152.Helbing, D. (2001). Traffic and related self-driven many-particle systems, Rev. Mod. Phys.

73(4): 1067–1141.Ho, I. W., Leung, K. K., Polak, J. W. & Mangharam, R. (2007). Node connectivity in vehicular

ad hoc networks with structured mobility, 32nd IEEE Conference on Local ComputerNetworks, LCN 2007. pp. 635–642.

Hoogendoorn, S. & Bovy, P. (2001). Generic gas-kinetic traffic systems modeling withapplications to vehicular traffic flow.

IEEE Draft Standard IEEE P802.11p/D9.0 (July 2009). Technical report.IEEE Trial-Use Standard for Wireless Access in Vehicular Environments (WAVE) - Multi-channel

Operation (2006). IEEE Std 1609.4-2006 pp. c1–74.IEEE Trial-Use Standard for Wireless Access in Vehicular Environments (WAVE) - Networking

Services (Apr. 2007). IEEE Std 1609.3-2007 pp. c1–87.ITU-R (2007). Propagation by diffraction, Recommendation P.526, International

Telecommunication Union Radiocommunication Sector, Geneva.Jerbi, M., Marlier, P. & Senouci, S. M. (2007). Experimental assessment of V2V and

I2V communications, Proc. IEEE Internatonal Conference on Mobile Adhoc and SensorSystems (MASS 2007), pp. 1–6.

Jin, W. (2003). Kinematic wave models of network vehicular traffic. Ph.D. Dissertation, UCDavis.

Kaul, S., Ramachandran, K., Shankar, P., Oh, S., Gruteser, M., Seskar, I. & Nadeem, T. (2007).Effect of antenna placement and diversity on vehicular network communications,Proc. IEEE SECON., pp. 112–121.

Koberstein, J., Witt, S. & Luttenberger, N. (2009). Model complexity vs. better parametervalue estimation: comparing four topography-independent radio models, Simutools’09: Proceedings of the 2nd International Conference on Simulation Tools and Techniques,ICST (Institute for Computer Sciences, Social-Informatics and TelecommunicationsEngineering), ICST, Brussels, Belgium, Belgium, pp. 1–8.

Kotz, D., Newport, C., Gray, R. S., Liu, J., Yuan, Y. & Elliott, C. (2004). Experimental evaluationof wireless simulation assumptions, Proc. ACM MSWiM ’04, ACM, New York, NY,USA, pp. 78–82.

Lighthill, M. J. & Whitham, G. B. (1955). On Kinematic Waves. II. A Theory of Traffic Flow onLong Crowded Roads, Royal Society of London Proceedings Series A 229: 317–345.

Mangharam, R., Weller, D. S., Stancil, D. D., Rajkumar, R. & Parikh, J. S. (2005). Groovesim: atopography-accurate simulator for geographic routing in vehicular networks, VANET’05: Proceedings of the 2nd ACM international workshop on Vehicular ad hoc networks,ACM, New York, NY, USA, pp. 59–68.

Martinez, F. J., Toh, C. K., Cano, J.-C., Calafate, C. T. & Manzoni, P. (2009). A survey andcomparative study of simulators for vehicular ad hoc networks (VANETs), WirelessCommunications and Mobile Computing .

Matolak, D., Sen, I., Xiong, W. & Yaskoff, N. (2005). 5 GHz wireless channel characterizationfor vehicle to vehicle communications, Proc. IEEE Military Communications Conference


www.intechopen.com


(MILCOM 2005), Vol. 5, pp. 3016–3022.Maurer, J., Fugen, T., Schafer, T. & Wiesbeck, W. (2004). A new inter-vehicle communications

(ivc) channel model, Vehicular Technology Conference, 2004. VTC2004-Fall. 2004 IEEE60th, Vol. 1, pp. 9–13 Vol. 1.

McCasland, W T (1965). Comparison of Two Techniques of Aerial Photography forApplication in Freeway Traffic Operations Studies, Photogrammetry and AerialSurveys.

Murthy, C. S. R. & Manoj, B. (2004). Ad Hoc Wireless Networks: Architectures and Protocols,Prentice Hall PTR, Upper Saddle River, NJ, USA.

Nagel, K. & Schreckenberg, M. (1992). A cellular automaton model for freeway traffic, J. dePhysique 2: 2221.

Nagel, R. & Eichler, S. (2008). Efficient and realistic mobility and channel modeling for vanetscenarios using omnet++ and inet-framework, Simutools ’08: Proceedings of the 1stinternational conference on Simulation tools and techniques for communications, networksand systems & workshops, ICST, Brussels, Belgium, Belgium, pp. 1–8.

Naumov, V., Baumann, R. & Gross, T. (2006). An evaluation of inter-vehicle ad hocnetworks based on realistic vehicular traces, MobiHoc ’06: Proceedings of the 7th ACMinternational symposium on Mobile ad hoc networking and computing, New York, NY,USA, pp. 108–119.

Network Simulator 2 (n.d.).URL: http://www.isi.edu/nsnam/ns/

Open Street Map Project (n.d.).URL: http://www.openstreetmap.org

Otto, J. S., Bustamante, F. E. & Berry, R. A. (2009). Down the block and around the corner – theimpact of radio propagation on inter-vehicle wireless communication, Proc. of IEEEInternational Conference on Distributed Computing Systems (ICDCS).

Palazzi, C. E., Ferretti, S., Roccetti, M., Pau, G. & Gerla, M. (2007). How do you quicklychoreograph inter-vehicular communications? a fast vehicle-to-vehicle multihopbroadcast algorithm, explained, in Proc. of the 3rd IEEE CCNC International Workshopon Networking Issues in Multimedia Entertainment (CCNC/NIME 2007), Las Vegas, NV,USA, IEEE Communications Society.

Parsons, J. D. (2000). The Mobile Radio Propagation Channel, John Wiley & Sons.Piorkowski, M., Raya, M., Lugo, A. L., Papadimitratos, P., Grossglauser, M. & Hubaux, J.-P.

(2008). Trans: realistic joint traffic and network simulator for vanets, SIGMOBILEMob. Comput. Commun. Rev. 12(1): 31–33.

Rappaport, T. S. (1996). Wireless Communications: Principles and Practice, Prentice Hall.Rothery, R. W. (1992). Car following models, In Trac Flow Theory .Saha, A. K. & Johnson, D. B. (2004). Modeling mobility for vehicular ad-hoc networks, in K. P.

Laberteaux, R. Sengupta, C.-N. Chuah & D. Jiang (eds), Vehicular Ad Hoc Networks,ACM, pp. 91–92.

Standard Specification for Telecommunications and Information Exchange Between Roadside andVehicle Systems - 5GHz Band Dedicated Short Range Communications (DSRC) MediumAccess Control (MAC) and Physical Layer (PHY) Specifications (Sep. 2003). ASTME2213-03 .

SUMO - Simulation of Urban MObility (n.d.).URL: http://sumo.sourceforge.net

Takahashi, S., Kato, A., Sato, K. & Fujise, M. (2003). Distance dependence of path loss for


www.intechopen.com


millimeter wave inter-vehicle communications, Proc. IEEE 58th Vehicular TechnologyConference (VTC 2003-Fall), Vol. 1, pp. 26–30.

Tonguz, O. K. & Boban, M. (2010). Multiplayer games over vehicular ad hoc networks: A newapplication, Ad Hoc Networks 8(5): 531 – 543.

Tonguz, O. K., Viriyasitavat, W. & Bai, F. (2009). Modeling urban traffic: a cellular automataapproach, Comm. Mag. 47(5): 142–150.

Treiber, M., Hennecke, A. & Helbing, D. (2000). Congested traffic states in empiricalobservations and microscopic simulations, Phys. Rev. E 62(2): 1805–1824.

U.S. Census Bureau TIGER system database (n.d.).URL: http://www.census.gov/geo/www/tiger

Uzcategui, R. A. & Acosta-Marum, G. (2009). Wave: a tutorial, Comm. Mag. 47(5): 126–133.Vehicle Safety Communications Project, Final Report (2006). Technical Report DOT HS 810 591, U.S.

Department of Transportation, NHTSA, Crash Avoidance Metrics Partnership.Wang, S. Y., Chou, C. L., Huang, C. H., Hwang, C. C., Yang, Z. M., Chiou, C. C. & Lin, C. C.

(2003). The design and implementation of the nctuns 1.0 network simulator, ComputerNetworks 42(2): 175 – 197.

Wang, Z., Tameh, E. & Nix, A. (2004). Statistical peer-to-peer channel models for outdoorurban environments at 2 ghz and 5 ghz, Vehicular Technology Conference, 2004.VTC2004-Fall. 2004 IEEE 60th, Vol. 7, pp. 5101–5105 Vol. 7.

Wisitpongphan, N., Bai, F., Mudalige, P., Sadekar, V. & Tonguz, O. (Oct. 2007). Routingin Sparse Vehicular Ad Hoc Wireless Networks, IEEE Journal on Selected Areas inCommunications 25(8): 1538–1556.

Wu, H., Palekar, M., Fujimoto, R., Guensler, R., Hunter, M., Lee, J. & Ko, J. (2005). An empiricalstudy of short range communications for vehicles, Proc. of the 2nd ACM Internationalworkshop on Vehicular ad hoc networks, pp. 83–84.

Zang, Y., Stibor, L., Orfanos, G., Guo, S. & Reumerman, H.-J. (2005). An error modelfor inter-vehicle communications in highway scenarios at 5.9ghz, PE-WASUN ’05:Proceedings of the 2nd ACM international workshop on Performance evaluation of wirelessad hoc, sensor, and ubiquitous networks, ACM, New York, NY, USA, pp. 49–56.


www.intechopen.com

Mobile Ad-Hoc Networks: ApplicationsEdited by Prof. Xin Wang

ISBN 978-953-307-416-0Hard cover, 514 pagesPublisher InTechPublished online 30, January, 2011Published in print edition January, 2011

InTech EuropeUniversity Campus STeP Ri Slavka Krautzeka 83/A 51000 Rijeka, Croatia Phone: +385 (51) 770 447 Fax: +385 (51) 686 166www.intechopen.com

InTech ChinaUnit 405, Office Block, Hotel Equatorial Shanghai No.65, Yan An Road (West), Shanghai, 200040, China

Phone: +86-21-62489820 Fax: +86-21-62489821

Being infrastructure-less and without central administration control, wireless ad-hoc networking is playing amore and more important role in extending the coverage of traditional wireless infrastructure (cellularnetworks, wireless LAN, etc). This book includes state-of the-art techniques and solutions for wireless ad-hocnetworks. It focuses on the following topics in ad-hoc networks: vehicular ad-hoc networks, security andcaching, TCP in ad-hoc networks and emerging applications. It is targeted to provide network engineers andresearchers with design guidelines for large scale wireless ad hoc networks.

How to referenceIn order to correctly reference this scholarly work, feel free to copy and paste the following:

Mate Boban and Tiago T. V. Vinhoza (2011). Modeling and Simulation of Vehicular Networks: towardsRealistic and Efficient Models, Mobile Ad-Hoc Networks: Applications, Prof. Xin Wang (Ed.), ISBN: 978-953-307-416-0, InTech, Available from: http://www.intechopen.com/books/mobile-ad-hoc-networks-applications/modeling-and-simulation-of-vehicular-networks-towards-realistic-and-efficient-models

Date post:	07-Feb-2018
Category:	Documents
Upload:	duongdung
View:	213 times
Download:	0 times

Modeling and Simulation of Vehicular Networks: Towards ...cdn.intechweb.org/pdfs/12878.pdf · VANET...

Documents